Identifying and Validating GSTM5 as an Immunogenic Gene in Diabetic Foot Ulcer Using Bioinformatics and Machine Learning

Abstract

Background

A diabetic foot ulcer (DFU) is a serious, long-term condition associated with a significant risk of disability and mortality. However, research on its biomarkers is still limited. This study utilizes bioinformatics and machine learning methods to identify immune-related biomarkers for DFU and validates them through external datasets and animal experiments.

Methods

This study used bioinformatics and machine learning to analyze microarray data from the Gene Expression Omnibus (GEO) database to identify key genes associated with DFU. Animal experiments were conducted to validate these findings. This research employs the datasets GSE68183 and GSE80178 retrieved from the GEO database as the training dataset for building a gene machine learning model, and after conducting differential analysis on the data, this study used package glmnet and package e1071 to construct LASSO and SVM-RFE machine learning models, respectively. Subsequently, we validated the model using the training set and validation set (GSE134431). We conducted enrichment analysis, including GSEA and GSVA, on the model genes. We also performed immune functional analysis and immune-related analysis on the model genes. Finally, we conducted immunohistochemistry (IHC) validation on the model genes.

Results

This study identifies GSTM5 as a potential immune-related key target in DFU using machine learning and bioinformatics methods. Subsequent validation through external datasets and IHC experiments also confirms GSTM5 as a critical biomarker for DFU. The gene may be associated with T cells regulatory (Tregs) and T cells follicular helper, and it influences the NF-κB, GnRH, and MAPK signaling pathway.

Conclusion

This study identified and validated GSTM5 as a biomarker for DFU. This finding may potentially provide a target for immune therapy for DFU.

Keywords:

• bioinformatics
• machine learning
• diabetic foot ulcer
• LASSO
• SVM-RFE
• GSTM5

Data Sharing Statement

All generated data have been included in the published manuscript. The datasets of gene expression generated during the current study are available in the NCBI Gene Expression Omnibus (GEO) database (accession number: GSE68183, GSE80178, and GSE134431). For all other information, contact with the author (Hongshuo Shi, [email protected]).

Disclosure

The authors declare no competing interests.

Additional information

Funding

The study was funded by the National Natural Science Foundation of China (82274528); Construction Task Book for the Three-Year Action Plan for Accelerating the Inheritance and Innovative Development of Traditional Chinese Medicine in Shanghai (2021-2023) (ZY(2021-2023)-0211); Shanghai Municipal Health Commission Scientific Research Programme Mission Statement (202240228); Special Youth Project for Clinical Research of Shanghai Municipal Health Commission (20234Y0162); Clinical Research Talent Training Program of Shanghai University of Traditional Chinese Medicine Affiliated Hospital (2023LCRC06); Four Bright Foundations of Shuguang Hospital Affiliated to Shanghai University of Traditional Chinese Medicine (SGKJ-202301); Science and Technology Development Fund of Shanghai University of Traditional Chinese Medicine (Project No. 23KFL107).